A PG-LSH Similarity Search Method for Cloud Storage

LSH (Locality Sensitive Hashing) indexing is often used for similarity search in high-dimensional object set. The way of separating index and store in LSH is particularly suitable for searching encrypted data stored in the cloud. However, most LSH methods only use or improve the first phase of LSH for similarity search. In this paper, we focus on the second phase of LSH, and propose a probability guarantee LSH (PG-LSH) method based on the Multi-Probe method, and apply it to cloud storage encrypted data search. Our approach simplifies the candidate set generation process and removes pseudo-similar objects. Experimental results show that the elimination rate of PG-LSH is more than 21%. It can also effectively reduce the time and space without losing search quality.